Professional Documents
Culture Documents
com
Department of Electrical and Computer Engineering, King Abdul Aziz University, P O Box:80204, Jeddah 21589
mamdohgharbi@yahoo.com
Abstract: The Moment Method (MM) is employed in the spectral domain where piece-wise sinusoidal basis
functions are used to expand the current distribution on the dipole. The current distribution and input impedance are
calculated as functions of the layers parameters. The effects of the material anisotropy and superstrate (cover)
parameters are studied. Expressions for the far-zone fields of the dipole in the anisotropic substrate-superstrate
structure are derived, and radiation patterns are computed for different structure parameter. Finally, Improvement of
Single-Element Antenna Design by using anisotropic layer will be demonstraited.
[Affandi A, Gharbi M, Ali S. Application of The Moment Method Spectral Integral Domain. Life Sci J
2015;12(5):1-23]. (ISSN:1097-8135). http://www.lifesciencesite.com. 1
Keywords: Moment method; dyadic Green’s function; current distribution; input impedance
1
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
j 0 0 1
r1TE =
1 RTE1 RTE1 e 2ik1 z
(h)
h1
j = 0 j 0 (2.8)
0 0 jz 1
(2.1) r1TM = (e )
TM
1 R 1 RTM1 e 2ik1 z h1 (2.9)
and 1
j 0 I
k (h)
jz = k 2
j k 2
s 2
(2.10)
j 2
k (jze ) = ( k 2j ks )
1
where
I
is the unit dyadic.
2
jz
The electric field integral equation EFIE on the (2.11)
surface S of the microstrip dipole can be written as: 2 2
k = 0 j
j
( s) (i) (2.12)
E t = E t , 1
(2.13)
surface of the printed dipole and
( s) T k 02 = 2 0 0 (2.14)
E t (r ) = i dr G 11 (r , r ) J s (r ) where j = 1,2.
S (2.2) The sommerfeld condition at infinity determines
T the branch interpretation of
G 11 ( r , r ) 1
where is the transverse 2x2 dyadic
k 0 z = k 02 k s2 2
as:
k jz
Green’s function which is written as below with 1
= k 2
k s2 2
k 0z 0 , k s k0
( h) k jz
replaced by k 1z for TE waves and replaced by
Re ( k 0 z ) 0 , and
k `1(ez) for TM waves, we get:
k s ( rs rs)
k 0 z = i k s2 k 02 , k s k0
G 11 (r , r ) = dk s e g11 ( k s , z)
z =0 i.e Im( k 0 z ) 0 .
TE
g11 (k s , z ) =
i0
8 2
r
k
1
(h)
1z
hˆ(k (h)
1z
(h) ( h) ( h)
)eik1 z z + RTE1 e 2ik1z h1 hˆ(k1(zh ) )e ik1 z z In the above Re ( k 0 z ) and Im( k 0 z ) stand for
real and imaginary parts respectively.
hˆ(k (h)
1z ) + RTEI hˆ(k1(zh ) ) That is the positive branch of the square root has
TM
+
r
k
1
(e )
vˆ(k ( e)
1z
( e) (e )
)eik1 z z + RTM1 e2ik1 z h1 vˆ(k1(ze ) )eik1 z
(e )
z
been properly chosen to be consistent with the
radiation condition as z .
1z
vˆ(k (e )
) + RTM1 vˆ( k1(ze ) )
1z (2.3) hˆ(k jz ) vˆ(k jz )
The unit vectors and are
where obtained as
k1(zh ) k 0 z 1
RTE1 = R10TE = (xˆky yˆkx )
k 1(zh) + k 0 z hˆ(k (h )
jz ) hˆ( k (hjz) ) k
(2.4) = = s = ĥ (2.15)
(h) (h) (h)
TE k ik2z 1z tan k h 2z 2 k (jze) j ks
R 1 = (h) (h) ( h) (e ) ( xˆk x + yˆ k y ) + zˆ
k ik tan k h vˆ( k ) jz k j ks jz k j
2z 1z 2z 2 (2.5) = =
k (e)
1 k0z j ks
uˆ (k (jze ) ) + zˆ
RTM1 = R10TM = 1z
(e) jz k j
k 1z + 1 k 0z (2.6)
(2.16)
where j = 1, 2.
2 k 1(ze ) i 1 k 2( ez ) tan k 2( ez ) h2 The expression for the transverse (surface)
RTM1 =
2 k 1(ze) i 1 k 2( ez ) tan k 2( ez ) h2 (2.7)
g 11 (k s , z = 0)
components of
Is
2
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
N
i 0 r1TE
g 11T (k s , z = 0) =
8 2 k
TE 2 ik ( h ) h1
( h) 1 + R1 e 1 z 1 + R I hˆhˆ
TE
J s (rs) = xˆ I n f n ( x , y )
1z n =1 (A/m) (2.22)
TM
+
r
k
1
( e)
1 R TM
1 e 2ik1( ze ) h1
1 R TM
1 uˆ (k1(ze) )uˆ (k1(ze) )
where
f n ( x , y ) = n ( x ) n ( y ) (2.23)
1z
(2.17) For the present work, Piecewise sinusoidal
Since the induced surface current density on the (PWS) expansion modes are selected to represent the
dipole is assumed to be in the x-direction, only the variation of the current in the x-direction.
The form of the nth expansion function centered
(x,x) component of G 11 will be required. Thus, about xn may be expressed mathematically as:
2
i k y r TE sin k ( x xn -1 )
, xn-1 x xn
(h )
g11( x , x ) ( k s ) = 02 2 1( h) 1 + RTE1 e 2ik1z h1 1 + RTE1
8 k s k 1z
sin kd
sin k ( x x)
n +1
, xn x xn +1 (2.24)
2 ( e)
( x ) =
k k
(e )
+ x2 1z2 r1TM 1 RTM1 e 2ik1z h1 1 RTM1 xˆ xˆ
k s k1
sin kd
(2.18)
Substituting Eq. 2.3 in Eq. 2.2 with z = 0, 0 , otherwise
E t( s) (rs ) = i dr dk
s s e ik s rs g11( x , x ) ( k s ) e iks rs J s rs where xn = nd, and d is the half length of each
S
expansion mode given by d = L / N.
(2.19)
The wavenumber k for the PWS mode can be
Using the 2-D Fourier Transform, Eq 2.19
chosen arbitrary, however, a judicious choice will
reduces to:
improve the convergence. For single isotropic
substrate, it was found that setting k equal to the
E t( s ) (rs ) = i dk s e ik s rs g11( x , x ) ( k s ) J s k s “effective” wave number for the substrate is a good
(2.20) choice (Pozar, 1983)
where J s ( k s ) is the Fourier Transform of the k = ke = e k0 (2.25)
surface current density J s ( rs ) .using electric field The y-variation of the current density may be
integral equation (EFIE) for the unknown surface chosen as
current density J s
1 W W
n ( y ) = , y
E t
( s)
= Et
(i)
, W 2 (2.26)2
and Thus, The Fourier transform of the current
getting
density
J s ( x, y ) in Eq. 2.26 can be obtained as
i
ik r ( x, x ) (i )
dk s e s s g11 ( k s ) J s k s E (rs ) , rs S N
(2.21a) ~
J s (k s ) = xˆ I n Fn (k s )
n =1 (2.27)
dk e s
ik s rs
J s k s 0, rs S where
(2.21b) ~ ~ ~
Eqs. 2.21 are the coupled set vector integral
Fn ( k s ) = n ( k x ) n ( k y )
(2.28)
equations in the spectral domain. Eq. 2.21a is the and
spectral domain electric field integral equation ~ 2 k e (cos k e d cos k x d ) ik x nd
(SDEFEI). n (k x ) = e
sin k e d ( k x2 k e2 )
2.1. Application of the Moment Method (2.29)
(MM)Solution W
To apply the MM to solve Eq. 2.21, the surface sin k y
current density on the dipole is expanded in terms of
~ 2
n ( k y ) =
f n ( x , y) W
appropriate expansion modes with ky
unknown coefficients In: 2 (2.30)
Substituting Eq. 2.27 in Eq. 2.21, we get
3
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
N
VmP = E
( s)
( x , y , z) J (i ) ( x , y , z) dx dy dz
I n E n( s ) (rs ) = E (i ) (rs ) , on S m
V (2.37b)
n=1 (2.31) Substituting Eq. 2.32 with n = m in Eq. 2.36 and
where E n ( rs ) is the scattered electric field due performing the integration over Sn, we get:
~ ~
to the n expansion mode f n of the surface current
th Zmn = i dk x dk y Fm ( k x , k y ) g11( x , x ) ( k x , k y ) Fn ( k x , k y )
- ,
density J s , and is given by m = 1,2,..., N
(s) ~ n = 1,2,..., N
E n (rs ) = xˆi dk s e ik s rs g11( x , x ) (k s ) Fn (k s ) (2.38)
- The system of equations given in Eq. 2.35 can be
n = 1,2,..,m,...,N (2.32) written in matrix form as
where g
( x ,x)
( k s ) is given by Eq. 2.8. Applying
11
[ Zmn ] [ In ] = m
V P
(2.39)
the Galerkin’s method by choosing the test function to where [ I ] is the generalized current column
be the same as the expansion functions; dot vector whose N components contain the In of Eq. 2.22,
multiplying Eq. 2.27 by f m ( x, y ) and integrating [ Zmn ] is the NxN generalized impedance matrix, and
over the surface of the dipole, we get
N
V is the generalized voltage column vector. The
m
P
n 1 Sm
amperes (VA), while the elements In are
f m ( x , y ) E (i ) ( x , y ) dx dy dimensionless.
Sm
= , m= 1, 2,.., N 2.2 Calculation of the Generalized Voltage Matrix
(2.33)
where Sm is the surface of the mth test mode and V m
P
E m( s) ( x, y ) is the scattered field in layer (1) by the mth test mode Iff mtheatdipole is excited by a delta-gap generator,
the position of the nth expansion mode. P
For the case of impressed current source (probe the evaluation of the voltage vector elements Vm
excitation), by using the reciprocity theorem, Eq. 2.33 given by Eqs. 2.37 is simple: only these position in the
can be written as generalized voltage matrix corresponding to segment
N junction that contain the generator will have non zero
( s)
I n E ( x , y ) f n ( x , y ) dx dy P
value (Vm =1 for a delta-gap generator of 1V) (Lo et
m
n 1 Sn
( s)
al., 1977).
E m ( x, y, z) J (i ) ( x, y, z) dx dy dz When the dipole is excited by a probe, the
P
= V , evaluation of Vm is rather involved. In this case, we
m = 1, 2,.., N (2.34)
( s) need to find E m ( x , y , z ) which is the electric field in
where, E m ( x , y , z ) is the field in layer(2) of the dielectric layer (2) due to the mth expansion mode
J (i ) f m ( x , y ) of the surface current density on the
the test mode at the position of the probe and is
the impressed (source) current of the probe. dipole.
Thus, Eq. 2.33 or Eq. 2.34 can be written as: If the impressed source is assumed to be an
N idealized current source (probe feed) expressed as:
Z mn I n VmP J (i ) ( x, y, z) = zˆ ( x x p ) ( y y p )
n 1 , m = 1, 2,...., N (2.35) (2.40)
where where the feed position is at (xp,yp). The voltage
Z mn = E m( s ) ( x , y ) f n ( x , y ) dx dy vector elements reduce to
Sn
(2.36) V mP = E m ( x, y , z ) zˆ ( x x p ) ( y y p ) dV
0
VmP = f m ( x , y) E (i ) ( x, y) dV
V (2.37a)
=
- h2
E zm ( x p , y p , z ) dz
or (2.41)
4
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
( z ,x)
where Ezm is the z-component of E m . Using Eq. 2.44, Eq. 2.15 and Eq. 2.16 g 21
Using the dyadic Green’s function formulation, can be obtained as
we have ( z ,x)
g 21 ( k s , z) =
E m ( x , y, z) = i G 21 (r , rs) f m ( rs) dS TM
i 0 2 r
Sm
(2.42)
8 2 2 z k1 k 2
1
(e)
X TM2 ,1 k x 1 RTM1 e 2 ik1z h1
G 21 (r , rs)
where is the dyadic Green’s
function in layer (2) due to a point source in layer (1) e ik2( ez) z
+ RTM2 e 2ik2 z
(e )
h2
eik2 z
(e )
z
(2.48)
on the dipole surface. where the TE- components do not contribute.
From Eq. 2.43, Eq. 2.47 and Eq. 2.48, the
G 21 (r , rs)
for the layered uniaxially
expression for
E zm reduces to:
anisotropic medium can be given as:
E zm ( x p , y p , z )
G 21 (r , r ) =
z = 0
= i ( kx x p k y y p )
i dx dy dk s e
dk s e ik s ( rs rs) g21 ( k s , z) Sm -
(z,x)
- (2.43) g 21 ( k s , z ) e ik rs f m ( rs) (2.49)
where Since
g 21 ( k s , z ) = ik s rs ~
i 0 r1TE TE ˆ dx dy e f m (rs) = Fm ( k s )
2 (h)
8 k1z
(h )
(h ) ( h)
X 2,1 h k2( hz ) eik2 z z + RTE2 e 2ik2 z h2 hˆ k 2(hz ) eik2 z z Sm
(2.50)
thus
hˆ k ( h)
1z + R TE 2ik1(zh ) h1
1 e
hˆ k1(zh ) Ezm(xp, yp ,z)
=
+
i(k x x p k y y p )
r1TM TM i
k1(ze)
(e)
(e) (e)
X 2,1 vˆ k 2( ez ) e ik2 z z + RTM2 e 2 ik2 z h2 vˆ k 2( ez) e ik2 z z dk
-
x dk y e
~
vˆ k (e)
1z + R TM
1 e 2 ik 2( ez ) h1
vˆ k1(ze ) (2.44)
g ( z ,x)
21 ( k , z ) Fm ( k s )
s (2.51)
Substituting Eq. 2.51 in Eq. 2.41, we arrive at
where
X TE
=
1 + R TE
1 i dk x dk y e
i(k x x p k y y p )
2 ,1 VmP
1 + R TE
2
(h)
e 2ik2 z h2 (2.45) ~ ( z ,x )
Fm (k x , k y ) g21(
= -
I ) ( k x , k y ,z)
,
X TM k
= ( e2)
k ( e)
1z
1 R TM
1
m=1,2,....,N
2 ,1
k2z k1 1 R TM
2
(e)
e 2ik2 z h2 where
(2.46) 0
TE TM ( z, x)
R 2 = 1 R 2 1
g ( z,x)
21( I )
g
- h2
21 ( k x , k y , z) dz
=
Since f m (x,y) is assumed to be x-directed in Eq 2 r1TM k 1(ze )
(e )
=
8 2z k 2z
(2.52)
component of the electric field Ezm. Thus,
P
Ezm ( x p , y p , z) The Zmn and V appearing in Eq. 2.38 and Eq.m
=
2.51, respectively have to be computed using
i G21( z , x) ( x p , y p , z, rs) f m (rs)dS (2.47) numerical integration techniques and properly
Sm accounting for the residue contribution of the surface
where the (z,x) component only of the dyadic wave poles. The unknown complex coefficients In are
( z ,x) determined after solving Eq. 2.39.
Green’s function G 21 is required.
2.3 Calculation of Input Impedance
5
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
For the case when the dipole is excited by delta- The integrals in Eq. 2.57 and Eq. 2.58 are to be
gap generator, the input impedance may be found evaluated numerically. Thus, the expressions for Zmn
Vin and Vm constitute the bulk of the computational effort
Zin for the microstrip dipole solution, and so it is critical
from
I in . So, if
Vin 1 , the input that these terms be calculated as efficiently as
1 possible. The singularities of the integrand correspond
Z in to TM and TE surface waves. If the dielectric layers
impedance is simply I in .
are lossless, then the surface wave poles are on the
For a dipole excited by a probe, the input
impedance can be calculated as: j
N
Re{k} axis between k0 and kj, where kj= k0 is
I n VnP the wave number in the dielectric layer with higher
Zin = n=1 (2.53) dielectric constant. As illustrated in Figure 2.2, if the
where In are the expansion mode current dielectrics are lossy, then the poles move in the third
amplitudes and I is the probe current. quadrant. However, for low-loss dielectrics, the poles
2.4 Computational Efficiency are very close to the Re{k} axis. These poles must be
Because of the computational complexity of Eq. avoided in performing the numerical integration from
instead of performing the doubly infinite integrals in k=0 to. The conventional method of avoiding the poles
the (kx,ky) space, it is common to change to polar is to deform the contour from the Re{k} axis to the
coordinates (k) in the spectral domain, where contour C1 which makes a small half-circle around
kx = k cos (2.54) each pole.
ky = k sin (2.55) The integrals around the half-circle can be done
In this case, the integrations in Eq. 2.38 and Eq. in closed form by computing the residues at the poles.
2.52 undergo a transformation which can be This method works well for one layer structure and
symbolically shown as when the substrate is electrically thin, there is only
one surface wave pole (TM surface wave pole).
However, for multilayered structure (or if the substrate
U (k
-
x , k y ) dk x dk y
is not electrically thin), there can be several surface
=
2p
wave poles. Also, analytically performing the
integrations around the half-circles will be
d dk k U ( k , ) complicated if two or more poles are extremely close
0 0 (2.56) to each other. These problems can be entirely avoided
Thus, the doubly infinite integrals are reduced to by deforming the contour C1 into the contour C2.
a finite integration and a semi-infinite integration. The contour C2 goes down from the origin at 45
Also, the even and odd properties of the integrands angle until it is at height -jH below the Re{k} axis. It
can be used to reduce the = 0 2 integration is then proceeded parallel to the Re{k} axis until it is
past kj. Finally, it moves back up to the Re{k} axis
=0 and then out to . The advantage of the contour C2 is
range to 2 . Eq 2.38 and Eq. 2.52 can then that numerical integration near the poles is avoided.
be written as Further, no knowledge of the pole locations or their
/2 number is required. This method works well for H on
Z mn = 4i d dk k g
( x ,x )
11 k , the order of 0.1k0 (Derneryd, 1977, Carver, 1979). If
the dipole is divided into equal-length segments, all
0 0
~ ~ ~ ~ the values of the N2 matrix elements are contained in
Re n m Re n m any one row of [ Zmn], say the first one.
m = 1,2,..., N
n = 1,2,..., N (2.57)
and
/2
Vm = 4 d dk k
( z ,x )
g 21
k ,
0 0
~ ik x ~ ik y
Im{ m e x p } Re{n e y p } ,
Figure 2.2. Two paths of integration in the complex
k-plane
m=1,2,..., N (2.58)
6
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
All other rows are morally a rearranged version Thus, to calculate the far-field expressions we
of the first. The remaining elements can be obtained need to calculate E0z and H0z in the upper half space
by the rearrangement algorithm (Lo et al., 1977)
Zmn = Z1,m-n+1, m 2, n 1, (2.59) (layer(0)) due to the current distribution J s ( r ) =
Such a matrix is said to be a toeplitz matrix. In xˆ J x ( x , y ) on the microstrip dipole in layer (1).
this case execution time is considerably reduced from: The electric field E0z can be obtained as
t A N2 + BN3 (2.60)
zˆ E 0 ẑ
to E0z= = i
t A N + BN5/3 (2.61)
where A is the time required to compute a typical dr G 01 ( r , r ) J s ( r )
impedance matrix element, B N3 is the time required S'
to solve [ Zmn] [ In] = [ Vm] for [ In] by matrix inversion ( z , x)
for a system of order N, where [ Zmn] is a non-teoplitz dx dy G 01 (r , r ) J x ( x, y)
matrix. This time is reduced to BN5/3 by using a =i S' (2.67)
computer program for solving the teoplitz matrix. where
2.5 Radiation Field Evaluation G01(z , x ) ( r , r ) = zˆ G01 (r , r ) xˆ
The current distribution is radiating in the (2.68)
presence of the layered anisotropic medium. To find G01 (r , r ) can be obtained as (See Figure. 2.3).
the field patterns, the expression for electric and
magnetic field is obtained in the upper half-space. and r in region (1) where z h1
Using the dyadic Green's function in the upper half-
space due to the current distribution on the dipole. G01(z , x ) ( r , r ) =
These fields are represented in terms of Sommerfeld-
Weyl-type integrals. These integrals are calculated in
dk x dk y eik s ( rs rs) g01
( z,x )
( k s , z, z )
the radiation zone or far-field region. Such integrals - (2.69)
can be evaluated with numerical integration routines,
but it becomes very tedious when observation point is
in the far field of the source, because the integrand is
rapidly oscillating. However, in the far-field region,
we can use the stationery phase method or the saddle
point method to evaluate the integral to obtain far-
field expressions. Here a method that captures the
essence of the stationary phase method is presented
and the far field approximations of the Sommerfeld-
weyl type integrals are derived. The result of this
calculation is correct asymptotically to the leading-
order which we are interested in.
2.6 The Electric and Magnetic Fields in the Upper-
Half-space Figure 2.3, Geometry of problem with sources in
In the upper-half space (region 0), the far-field region(1) and observation point in region (0).
satisfies the following relations:
E0 z For the case of uniaxially anisotropic medium,
E 0 = G01(z , x ) ( r , r )
sin (2.62) =
1
E 0 dk x dk y e iks ( rs rs) g01
( z ,x )
( k s , z, z )
H 0 0 -
= (2.63)
for TM waves, and where
H0z i 0 r1TM TM
H0 = sin ( z ,x ) (e)
X 0,1 zˆ
g 01 ( k s , z, z ) 2 k
(2.64) = 8 1 z
E 0 0 H0
= (2.65) vˆ(k )e 0z
ik 0 z z
7
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
(i)
hˆ k1(zh ) + RTE1 e ik1 z h1 hˆ k1(zh) }
(h)
and kx is taken outside the integral as x . r1TM TM ik1(ze ) h1
(e)
X 0,1 e (ik 0 )hˆ(k 0 z )
Thus, E0z is obtained and expressed in Eq. 2.75 in k1z
terms of Weyl-type integral. E0z can also be expressed +
in terms of a Sommerfeld-type integral as
vˆk (e)
1z + R TM
1
(e)
e ik1 z h1 vˆ k1(ze ) } (2.82)
(i) 0 dx dy J x ( x , y ) Hence, the magnetic field
H0z can be obtained
E0z = 4 S
as
cos dk k 2 g E ( k ) e ik0 z z J 1 k 1
0 (2.77) H0z= 0 zˆ H
=
0
where we used the relation
d x d y zˆ G 01 xˆ J x ( x , y )
J 0 k S (5.22)
x i
cos k J1 k - (2.78) 2
dx dy J x ( x , y )
H0z = 8 S'
8
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
( i ) dk s
y - d
I= f() h(,)
(2.91)
e
i k ( x x )+ k ( y y ) where f() is the slowly varying part, while
z z
e ik 0zz
g H (ks ) (2.83) h(,) is the rapidly varying part when .
where Assuming here that h(,) is of the form
r1TE h(,) = eis( (2.92)
X TE0,1 Next, finding the stationary phase point of
g H (k s ) =
k1(zh ) h(,) which is given by the value of where
e ik1( zh ) h1
1 + R TE
1 e ik1( zh ) h1
S( )
( )
(2.84) 0
=0 (2.93)
where Most of the contributions to the integral in Eq.
1 5.31 will be from the vicinity of the stationary phase
(h) point. Hence, replacing the slowly varying part of the
r1TE =
1 RTE1 RTE1 e 2ik1 z h1 (2.85) integrand f() by its value at the stationary phase
point. By so doing, Eq. 2.91 becomes
R TE1 R10TE
= (2.86)
=
X TE
1
0,1 1 + R TE
(2.87)
d
I ~ f() h(,),
(2.94)
Thus, H0z is obtained and expressed in terms of If the integral has a closed-form expression, Eq.
Weyl-type integral. In a similar way, H0z can be 2.94 is an analytic expression for the leading-order
expressed in terms of a Sommerfeld type integral as approximation of Eq.2.90. An illustration of how the
1 foregoing method is applied to find the far-field
dx dy J x ( x , y )
approximations of Sommerfeld-Weyl-type integrals.
H0z = 4 S
The following two identities are used in this
approximation:
sin dk k 2 g H ( k ) e ik0 Z z J 1 k 1) The Weyl Identity (Richards and Lom 1981):
0 so
(2.88)
e
i kz x+k y y+kz z
where the relation e ikr i
dk x dk y
J 0 k r = 2 - (2.95)
kz
y = 2) The Sommerfeld Identity (Richards and Lom
1981): thus
sin J 1 k
i k (1)
(2.89) e ikr dk H 0 k e ik z z
Can be written. 2 k
r = - z
(2.96)
2.7 Leading-Order Approximations to 2 2 2
kz k k x k y
Sommerfeld-Weyl-Type Integrals In Eq.2.95, = , and in
Let us consider an integral of the type 2 2
k k k
Eq. 2.96 z =
I
d g ( , ) Let as consider first a Weyl-type integral in the
form we encountered for the fields in the upper half-
(2.90)
where is a large parameter. If the integrand space as
becomes rapidly oscillating when is large, and it
there exists a simple stationary phase point in the dk x dk y g k x , k y
e
i kz x + k y y + k z z
integrand, we can evaluate a leading order I1 = - (2.97)
approximation of Eq. 5.30 using the method of where kz is defined in Eq. 2.95. When x, y, or z
stationary phase (Lo et al., 1979). are large, the integrand in Eq. 2.97 is rapidly
The first step in the procedure is to factorize the oscillating, and the integral is amenable to the
integrand g(,) into two parts, a slowly varying stationary phase approximation. With the Weyl
part and a rapidly varying part. So, we let identity in mind, factorizing the integrand into slowly
9
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
varying part and a rapidly varying part which can be Let us next consider a Sommerfeld-type integral
integrated in closed form; that is, writing of the form
dk
-
x
dk y g k x , k y k z dk k g k J
n
n ( k ) eikz z
I1 = I2 = (2.103)
e i k z x + k y y + k z z
Where kz is defined following Eq. 2.96. When
, using the large argument approximation for
kz
(2.98) the Bessel function, notice that the integrand does not
By virtue of the stationary phase method have a well-defined stationary phase point. To obtain
argument, contributions to the integral will come from an integrand with the form of Eq. 2.96, it is most
around the point where the phase of the exponential appropriate to convert the above integral from a semi-
function is stationary. Hence, looking for the infinite integral to an infinite integral and replace the
stationary phase point in the two-dimensional kx-ky Bessel function with a Hankel function. This will lead
space which is given by to the use of the Sommerfeld identity given in Eq.
2.96.
Before proceeding any further, notice that Eq.
k x i k x x + k y y + k z z =0, 2.103 is also obtained when writing Eq. 2.97 in terms
of cylindrical coordinates by letting
kx = k cos, ky = k sin, x = cos, y = sin (2.104)
k y i k x x + k y y + k z z Then Eq. 2.97 becomes
=0 (2.99) 2
This is easily solvable to give the stationary
phase point at dk k d gˆ k ,
0 0
x y I1=
ik cos( ) ik z z
kx0 = k r , ky0 = k r (2.100)
e (2.105)
g (k
x2 + y2 + z2 where ) = g(k cos k sin).
where r = . Eq. 2.97 or Eq.
2.98 can be physically interpreted as expansions in a Next,
g (k) can be expanded in Fourier series in
spectrum of plane waves. The physical interpretation the -variable as:
of the stationary phase point, as expressed by Eq.
k -vector e in
gˆ n k
2.100, is that only the plane wave whose g (k
)=
n =- (2.106)
k = xˆk x + yˆk y + zˆk z
points from the source Substituting back Eq. 2.106 into Eq. 2.105 and
point to the observation point is important in making use of the following integral identity for the
evaluating the field at the observation point. By Bessel function
2
replacing the slowly varying part of the integrand in 1
Eq. 2.98 with its value at the stationary phase point, d ik cos in in /2
getting: Jn(k)= 2 0 e (2.107)
i ( k xx +k y y+kz z ) Obtaining
dk x dk y e
I1 ~ g(kx0,ky0) kz0 , r
kz
2 e in +in /2
dk k gˆ k
n
(2.101) I1 = n =- 0 Jn(k)
Using Weyl identity, Eq. 2.101 can be evaluated ikz z
in closed form to obtain the leading-order e (2.108)
approximation to the integral as Now identity each term in the series to be similar
e ikr to Eq. 2.103. From Eq. 2.106, if we denote (kx,ky) by
(k) in cylindrical coordinates, then the point (
I1 ~ 2 i g(kx0, ky0) kz0 r ,r (2.102) k x , k y k
where ) can be denoted by (k) or ( ). Because
of this, deduce that
k 0z k 2 k x20 k y20 k z / r ĝ ( k )= ein ĝ
. n n(k) (2.109)
10
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
i.e. Jn(x) = 2 n n
1
H ( 1)
( x) H (2) ( x ) the stationary point and use Sommerfeld identity to
evaluate the rapidly varying part in closed form.
(Abramowitz and Stegun, 1986), writing Eq. 2.103 as As a result
H n(1) ( k 0 )
1
2 dk k g k
n
H (1)
k e
ikz z
H
i g n k 0 (1)
(k 0 )
k z0 e ikr
I2 = 0 n
I2~ 0 r ,
r
1
2 dk k g k n
H n( 2 ) k eikz z
(2.116)
+ 0 (2.110) When r , using the large argument
k approximation of the Hankel function to simplify
With the change of variables from k to for
the second integral, the use of the property of the
H n(1) k
Hankel function that H 0(1) k in /2 k
~e , (2.117)
H (2)
n
x e = e
i in
H ( 1)
n ( x ) [13], and the Hence Eq. 2.116 is simplified to
property for gn(k) given by Eq. 2.109, combining the e ikr
two integrals in Eq. 2.110 into one integral as
I2 ~
( i ) n 1
gn k 0 k z0 r , r
dk k g k
1 (2.118)
I2 =
2
n
n
e H (1) k ikz z
2.8 Far-Field Expressions
(2.111) The results obtained in section Eq.2.63 are
The integrand in Eq. 2.111 is rapidly oscillating applied to find the far-field expressions.
when or z tends to because The z- component of the electric field in the
2 upper half-space given by Eq. 2.77, can be written as
H (1)
n x ~ x ei xn /2- /4 i 0
(2.112)
when x (Newman and Tulyathan, 1981). E0z( r ) = 4
Next factorizing the integrand in Eq. 2.111 into two
parts so that Sommerfeld identity can be used in Eq. dx dy J x ( x , y ) cos I E
2.96. Consequently, S
H ( k ) (1) (2.119)
1
dk gn k H0(1) ( k ) k z
2
n
where
I2 =
k (1) ik z z
dk k
2
gE k
ik 0 z z
J1 k
k H0 k k z e
z
IE =
(2.120)
0 e
(2.113)
In view of Eq. 2.112, the first factor in Eq. 2.113 which can be put in the form
is slowly varying while the second factor is rapidly
varying when r (i.e. when the observation point
1
2 dk k
2
gE k
is far from the field point). Since the rapidly varying IE= e ik0 z z
part of the integrand is asymptotic to the phase factor
ik ik z z
H1(1) k
(2.121)
e , the location the stationary phase point is Using the large argument approximation for the
given by Hankel function, the rapidly varying part of the
integrand in Eq. 2.121 is asymptotic to the phase
i k k z z ik ik0 z z
k factor e . Thus, the location of the
k 0
=0 (2.114) stationary phase point is given by
The solution of Eq. 2.114 is
11
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
i k k 0z z dx dy J x ( x , y )
k k0
e ik0
(2.127)
=0 (2.122) Thus, using Eq. 2.62 - Eq.2.65, the field patterns
The solution is given by are readily obtained.
k 0
= k0 sin,
k0z0 = k0 3. Results
cos Numerical results of the problem of a microstrip
dipole embedded in a substrate-superstrate structure
(2.123) are presented. Numerical computations investigating
the effects of the thickness and the anisotropy of the
layers on the dipole current distribution and input
r impedance are performed. Also, the radiation patterns
with = sin-1 , (see Figure2.4). of the microstrip dipole printed in the considered
substrate-superstrate structures are calculated.
The numerical computations for the problem
under consideration using the dyadic Green’s function
(DGF) formulation where the moment method (MM)
is applied in spectral domain (SD). A microstrip
dipole of length L = 0.5 0 and width W = 0.0001 0
, on the top of an isotropic substrate of permittivity
1 325
. 0 and thickness
h1 01016
. 0.
12
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
z
( 1)
of the substrate is to decrease the resonant
length relative to the isotropic substrate. Also, the
impedance level of the printed dipole on the
anisotropic substrate is lower than the dipole on an
isotropic substrate. Finally, Figures 3.3(a) and 3.3(b)
show, respectively, the E - and H - radiation patterns
of the printed dipole. For this considered case, there is
no significant differences between the isotropic and
anisotropic substrates. However, the H - plane Figure 3.3(b), H-plane radiation plot of a printed
radiation plot shows a slightly lower directivity for the microstrip dipole on a single layer using delta gap
printed dipole on the anisotropic substrate. generator with W/2 = 0.0001l0, h =0.08l0.
13
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
14
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
Epsilam-10 material having is less directive while the H- plane is more directive
x 130
. 0 , z 102
. 0. than the isotropic case).
Figure. 3.7(a) shows the real part of the dipole’s
current distribution for a printed dipole on an
anisotropic substrate and having isotropic superstrate.
Three different substrates are considered; where
z 10.2 z
17
. , 1
x 6 x (isotropic case), and
z 10.2
0.7846
x 13 . The figure shows that the
anisotropy ratio has great influence on the real part of
the current distribution. Substrates with negative
uniaxial material has the largest variation of the real
part of the current distribution, while substrates with Figure 3.7(a), Real part of current distribution of a
positive uniaxial material has the smallest variation. printed microstrip dipole using delta gap generator
Figure 3.7(b) shows that the imaginary part of the with W/2 = 0.0001l0 , L= 0.5 l0, h1 = h2 =
dipole’s current distribution does not depend too much 0.08 l0,e1 =3.25 e0 I,ez2 = 10.2 e0
on the anisotropy ratio. However, the current level is
decreased for the case of positive uniaxial material.
Fig. 3.7(c) shows that the magnitude of the
dipole’s current distribution is the highest for
substrates with negative uniaxial material. Figures
3.8(a) and 3.8(b) show the real and imaginary parts,
respectively, of the input impedance of the printed
microstrip dipole, as function of the electrical length
of the dipole for different values of the anisotropy
ratios. It is observed that as the anisotropy ratio (AR)
decreases, the resonant length of the dipole decreases.
Also, the magnitude of the real part of the input
impedance decreases with the decrease of the AR. The
figures also show that the shift in resonant frequency
is about equal to the bandwidth of the microstrip
dipole. This implies that a dipole designed for a Figure 3.7(b), Imaginary part of current distribution of
specific operating frequency may actually resonant a printed microstrip dipole using delta gap generator
outside the dipole bandwidth if the substrate with W/2 = 0.0001l0, L= 0.5 l0, h1 = h2 = 0.08 l0, e1
anisotropy is neglected (not considered). Similar =3.25 e0 I, ez2 = 10.2 e0
conclusion was obtained (Mosig and Gardiol, 1979),
for microstrip antennas on single anisotropic substrate.
Figures 3.11(a) and 3.11(b) show, respectively, the E-
and H- plane radiation patterns for three cases:
isotropic substrate with
2 z2 10.2 0 ,
positive uniaxial substrate with
2 6 0 , z2 10.2 0 , and negative uniaxial
substrate with 2
13 0 , z 2 10.2 0 .In these
Figure 3.7(c), Magnitude of the current distribution of
a printed microstrip dipole using delta gap generator
three cases the superstrate is assumed to be isotropic. with W/2 = 0.0001l0 , L= 0.5 l0, h 1 = h2 =
The figures show that for the positive uniaxial 008 l0, e1 =3.25 e0 I, ez2 = 10.2 e0
substrate, the E-plane radiation pattern has higher
directivity than the isotropic case while the H- plane
3.4 Effect of cover-layer anisotropy
radiation pattern has lower directivity. The radiation Figures. 3.10(a)- 3.10(c) show, respectively,
patterns of the printed circuit dipole on a negative the real part, imaginary part, and the magnitude of the
uniaxial substrate exhibits opposite property (E-plane current distribution of a printed dipole in a two-layer
15
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
16
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
17
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
18
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
19
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
20
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
21
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
Figure 3.40. Farfield realized gain plot for patch The Improved design (3) parameters for the
antenna antenna are detailed below:
For layer 1
The Improved design (1) parameters for the L1= 10.40 mm(Antenna length)
antenna are detailed below: W1= 25.4 mm (Antenna Width)
For layer 1 H1= 1.48 mm (Substrate Height)
L1= 10.40 mm(Antenna length) ∈ ( , , ) = (13,13,10.2)
W1= 25.4 mm (Antenna Width) Mt=.038 mm (Metal Thickness)
H1= 1.48 mm (Substrate Height) For layer 2
∈ ( , , ) = (6,6,10.2) L2= 10.40 mm(Antenna length)
22
Life Science Journal 2015;12(4) http://www.lifesciencesite.com
References
1. Abramowitz M. and I. A. Stegun,” Handbook of
Mathematical Functions,” New York: Wiley Interscience,
1986.
2. Byron E. V.," A New Flush-Mounted Antenna Element for
Phased Array Application," in Proc. Phased Array Antennas
Symposium, pp. 187-192, 1970.
3. Carver K. R.," A Model Expansion Theory for The
Microstrip Antenna," Dig. Int. Sump. Antenna Propagation
Soc., Seattle, WA. pp. 101-104 June 1979.
Figure 3.43 Farfield realized gain plot for patch 4. Carver K. R. and J. W. Mink," Microstrip Antenna
antenna (∈ ( , , ) = (13,13,10.2)) Technology," IEEE Trans. on Antennas and Propagation,
Vol. AP-29a, NO. 1, pp 2-24, January 1981.
5. Derneryd A. G.," A Theoretical Investigation of the
The Improved design (4) parameters for the Rectangular Microstrip Antenna Element," Rome Air
antenna are detailed below: Development Center, Tech. Rpt.TR-77-206, June 1977.
For layer 1 6. Deschamps G. A., "Microstrip Microwave Antennas."
presented at 3rd USAF Symposium on Antennas, 1953.
L1= 10.40 mm(Antenna length) 7. Gurel C. S. and E. Yazgan., “Characteristics of a circular
W1= 25.4 mm (Antenna Width) patch microstrip antenna on uniaxially anisotropic substrate,”
H1= 1.48 mm (Substrate Height) IEEE Trans. Antennas Propag., vol. 52, no.. 10, pp. 2532–
2537, Oct. 2004.
∈ ( , , ) = (13,13,10.2) 8. Gutton H. and G. Baissinot, "Flat Aerial for Ultra High
Mt=.038 mm (Metal Thickness) Frequencies," French Patent No. 703113, 1955.
For layer 2 9. Lo Y. T., D. D. Harrison, D. Solomon, G. A. Deschamps and
F. R. Ore," Study of Microstrip Antennas, Microstrip Phased
L2= 10.40 mm(Antenna length) Arrays, and Microstrip Feed Networks," Rome Air
W2= 25.4 mm (Antenna Width) Development Center, Tech Rpt. TR-77-406, October 1977.
H2= 1.48 mm (Substrate Height) 10. Lo Y. T., D. Solomon and W. F. Richards," Theory and
Experiment on Microstip Antennas, " IEEE Trans. on
∈ ( , , ) = (18,18,10.2) Antennas and Propagation, Vol. AP-27, No. 2, pp. 137-145,
Mt=.038 mm (Metal Thickness) March 1979.
11. Mosig J. R. and F. E Gardiol," The Near Field of An Open
Microstrip Structure," IEEE AP – S International
Symposium, pp. 379 - 382, 1979.
12. Munson R. E.," Conformal Microstrip Antennas and
Microstrip Phased Arrays," IEEE Trans. on Antennas and
Propagation, Vol. AP-22. No. 1, pp. 74-78, January 1974.
13. Newman E. H. and P. Tulyathan," Analysis of Microstrip
Antennas using Moment Methods," IEEE Trans, on Antennas
and Propagation," Vol. AP- 29, No. 1, pp. 47-53, January
1981.
14. Newman E. H.," Strip Antennas in a Dielectric Slab," IEEE
Trans. on Antennas and Propagation, Vol. AP-26, No. 5, pp
647-653, September 28, 1978.
Figure 4.28 Farfield realized gain plot for patch 15. Pozar D. M., " Considerations for Millimeter Wave printed
Antennas," IEEE Trans. on Antennas and Propagation, Vol.
antenna (∈ ( , , ) = (18,18,10.2)) AP-31, No. 5. pp. 740-747, September 1983.
16. Richards W. F., and Y. T. Lo, "An Improved Theory for
Acknowledgements: Microstrip Antennas and Applications, "IEEE Trans, on
Antennas and Propagation, Vol. AP-29, No. 1, pp. 38-46,
Authors are grateful to Mr. Mubsher for all January 1981.
support to carry out this work. 17. Uzunoglu N. K., N.G. Alexopoulos and J. G. Fikioris,"
Radiation Properties of Microstrip Dipoles," IEEE Trans. on
Corresponding Author: Antennas and Propagation, Vol. AP-27, No. 6, pp. 853- 858,
November 1979.
Mamdoh Gharbi
4/8/2015
23